Make or buy, when a casting beats a billet

The part is a pump housing. Roughly 4 by 4 by 3 inches finished, ductile iron, with a bored seat, 2 faced surfaces, a flange with 6 tapped holes and an internal passage. The customer asks for 400 in year one and says the program might reach 6,000 a year.

That last sentence is the whole quote. At 400 a year you machine it from solid. At 6,000 you cast it. Somewhere between those numbers is a crossover, and the make or buy decision is finding it rather than guessing. Every figure below is illustrative and one housing carries through.

The 2 routes a make or buy decision is choosing between

Machining from solid buys a block and removes most of it. For the housing, an illustrative rectangular billet of 4.5 by 4.5 by 3.5 inches weighs about 20 pounds in ductile iron, and the finished part weighs 6. Fourteen pounds becomes chips.

At an illustrative $1.40 a pound the block is $28, against a scrap recovery that returns very little. Then the machining is heavy, because removing 14 pounds takes roughing time, tooling and spindle hours, at an illustrative 2.4 hours across 2 setups.

The cast route buys a shape close to finished. A casting for this housing arrives at an illustrative $9.40 a piece at volume, weighs 7.5 pounds, and needs 0.7 hours of machining because only the seat, the faces and the tapped holes require metal removal.

The obvious conclusion is that casting wins, and it does on the piece price. What it carries instead is tooling, lead time and risk, which is where the decision actually gets made.

Tooling amortisation against material removal

A casting needs a pattern or a die, and that cost lands before the first part exists. For a housing at this size in sand, an illustrative pattern and core box costs $14,000. In a permanent mould or die process it is several times that and the piece price falls further.

Amortise $14,000 across 400 pieces and the tooling adds $35 a piece, which buries the casting route entirely. Across 6,000 it adds $2.33, and the casting route wins by a wide margin. Which volume you divide by is the tooling question in its own right.

The pattern decides everything.

That is the whole tension in a single line. The machined route front loads nothing and pays every year in material and spindle time. The cast route front loads the pattern and then pays very little.

Which is why the annual usage figure on the request matters more than any rate in the quote. A program quoted at 6,000 and released at 900 leaves the tooling unamortised and the supplier holding the difference, unless the quote said which volume it assumed.

Working out the crossover volume

The arithmetic is a single division and it beats any rule of thumb. Take the tooling cost, divide it by the per piece saving of the cast route, and the answer is the volume at which the 2 routes cost the same.

On the housing, the machined route is an illustrative $28 material plus 2.4 hours at $95, which is $228 plus $28, so $256. The cast route is $9.40 plus 0.7 hours at $95, which is $66.50 plus $9.40, so $76.

The saving is $180 a piece. Tooling of $14,000 divided by $180 is 78 pieces. Above roughly 78 pieces in total, the casting is cheaper even including the pattern.

At 400 a year the answer is not marginal, it is emphatic, and the instinct to machine low volumes from solid is wrong on this part because the material removal is so heavy. The instinct is right on parts where the billet is close to the finished shape and the saving per piece is small.

Divide, do not assert.

Run the same division on a bracket where the billet is nearly the finished shape and the saving is $6 a piece. Tooling of $14,000 divided by $6 is 2,333 pieces, and now the rule of thumb everybody carries is correct. The rule was never wrong, it was just being applied without the arithmetic that generates it, which is the same failure a machined part build-up exists to prevent.

Lead time, and what it costs commercially

A pattern takes weeks and the first castings take weeks after that. An illustrative 8 weeks for tooling and 4 for first samples means 12 weeks before a part exists, against a machined route that can ship in 2.

If the customer needs parts in 6 weeks, the crossover arithmetic is irrelevant, because one route cannot meet the date. The answer on programs like this is often both, machining the first lot from solid to hold the schedule while the pattern is made, then switching.

That hybrid has to be quoted as 2 prices with a stated changeover point, because a single blended price hides which lot costs what and leaves you carrying the difference if the volume never arrives.

Lead time also sits inside the supply chain risk. A casting supplier is another company with their own queue, their own minimum quantities and their own quality problems, which is the same exposure an assembly quote carries on every purchased line.

The quality risk that only the casting route carries

Castings arrive with porosity, inclusions and dimensional variation that solid material does not have. Porosity found after machining means the part is scrapped at its most expensive point, having absorbed the casting cost plus the machining.

That risk has a price and it belongs in the quote as a yield allowance rather than as optimism. An illustrative 3 percent casting scrap discovered after machining adds more than 3 percent to the cost, because each loss carries the machining that had already been done.

First article and process approval are heavier on the cast route too. The first pieces off a new pattern need dimensional verification across features that the pattern controls rather than features you machine, and a pattern correction is weeks rather than a program edit.

None of this makes casting the wrong answer. It makes the cast route a different risk profile at a lower cost, which is the trade the volume justifies above the crossover and does not justify below it.

Scrap late costs the machining too.

There is a machining consequence as well. A casting has a skin, and cutting through it is harder on tooling than cutting clean bar. Tool life on the first roughing pass into a sand casting is shorter, and if the casting supplier is inconsistent about stock allowance, the operator is taking a variable depth of cut on every piece.

What to state on the quote

Write down the volume you priced, the tooling assumption and who owns the pattern. Those 3 sentences prevent most of the arguments that follow an award on a part like this.

Pattern ownership is the one that gets skipped. If the customer pays for the pattern they will expect to take it elsewhere eventually. If you pay for it and amortise it in the piece price, the quote should say what happens if the program ends at 900 pieces rather than 6,000.

A customer running should-cost analysis on this part will model the cast route regardless of what you quote, because their model assumes the efficient process at the stated volume. Quoting the machined route at 6,000 pieces without explaining why invites a conversation you will lose.